rank-local
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides instructions and a workflow that involves processing external, untrusted data from customer reviews.
- Ingestion points: The skill suggests using the
seojuice:gbp-overviewtool (referenced in the Pro Tip section) to fetch business reviews, which are external inputs. - Boundary markers: Absent. The instructions do not include delimiters or specific warnings for the agent to ignore potentially malicious instructions embedded within customer review text.
- Capability inventory: The skill explicitly mentions that the agent can "reply to reviews directly," establishing a write-action capability based on untrusted external data.
- Sanitization: Absent. There are no instructions or scripts provided to sanitize, filter, or validate the content of reviews before the agent processes them or generates a response.
Audit Metadata